Pulitzer-prize winning poet William Carlos Williams wrote, “It’s not what you say that matters but the manner in which you say it; there lies the secret of the ages.” According to new research from Wharton marketing professor Jonah Berger, he was right. Berger and Alex Van Zant, management professor at Rutgers Business School, conducted four experiments on use of nonverbal communication in persuasion, finding that speakers who modulate their voices appear more confident, which makes them more likely to succeed in convincing their listeners to take action. Berger recently spoke with Knowledge@Wharton about their paper, “How the Voice Persuades,” which was published in the Journal of Personality and Social Psychology.

Knowledge@Wharton: What was the inspiration for this research?

Jonah Berger: Almost everyone has someone they want to persuade. Salespeople want to persuade the customer or the client. Marketers want to persuade the consumer. Leaders want to persuade employees. Managers want to persuade their boss. From politicians to folks in business, lots of people want to persuade others. But persuasion is often quite difficult. Often, when we try to persuade others, they’re less likely to do what we suggest. We wondered how might the voice that someone uses — the way they talk, in addition to the words they use — affects what other people do, affects whether other people listen?

Recently, I’ve been fortunate to do a lot of work on text analysis, which is processing textual data to understand how it impacts behavior — things like looking at the words used in customer service calls and how they impact satisfaction, or predicting the success of songs or movies based on their lyrics or their scripts. But while we looked at the language itself, the words themselves, we paid less attention to how those words were used. In this project, we wanted to begin to ask, “What about paralanguage? What about vocal features, and how might those aspects impact persuasion?”

Knowledge@Wharton: As you noted, sometimes the more you try to persuade someone, the less effective it is. What has past research said about why these attempts to persuade sometimes backfire?

Berger: There’s a concept in behavioral science called reactance, and I think this is something many of us are at least a little bit familiar with. When we try to push someone to do something, ask them to do something, persuade them to do something, they often don’t do what we want. They often do the opposite. They often push back. A good way to think about it is almost an anti-persuasion radar. It detects incoming projectiles and shoots up things to knock them down.

When people hear someone trying to persuade them, they push back. They react against the message. In some cases, we delete an incoming email if we know it’s trying to persuade us. We see an ad, we walk out of the room. But it’s not only that. We also counter-argue against those messages. If we watch an ad or listen to someone trying to persuade us, we’re often not just listening to them, we’re thinking about all the reasons that they’re wrong, why what they’re saying isn’t exactly true, how it could be different, why we shouldn’t listen to what they’re saying. For all those reasons, it’s really hard to persuade someone to do something. When they know we’re trying to persuade them, they’re actually less likely to listen to what we might want to say.

Knowledge@Wharton: Your research focuses on paralinguistic cues or how we say things, as opposed to the words that we’re speaking. What are some ways people use nonverbal communication in persuasion?

Berger: First, I think it’s important to talk about what we mean by paralinguistic cues, and that is all the vocal features from a conversation, from what someone might say to one another. How high-pitched or low-pitched someone is talking. I can talk at a very high pitch, and I can talk at a very low pitch. I can talk at a faster rate. I can speak very quickly, or I can speak more slowly. I can talk at a high volume, or I can talk at a lower volume. When you ask me a question, I can take a long time to respond, waiting for a couple of seconds to respond, or I can respond very quickly. All of those are examples of paralinguistic features, examples of ways that people might modulate their voice when talking to one another.

In this case, we found that when people are trying to persuade, they tend to use a number of different features. They increase their pitch. They vary that pitch more. They speak faster, in addition to raising their volume and varying their volume. They try a number of things when they try to persuade. Not all of them are effective, but when we’re trying to persuade, we tend to do a lot of things even without realizing that might shift the way our voice sounds.

Knowledge@Wharton: You studied this over the course of four experiments. What vocal cues did you find were the most effective?

Berger: We found a couple of things. First, we did a variety of experiments in different contexts. We asked people to try to persuade someone else to buy a television, for example, like they might in an online review. We asked people to dissuade others to do a task that we had asked them to do — to persuade them to do one task rather than another. In addition to changing what they might say, some participants also changed the way that they said it. Some participants tried really hard to persuade others, not through the exact words, but through what they said and how they said it. Others didn’t try so hard to persuade. They just used their regular voice.

What we found is that increasing and varying volumes — talking more loudly but also varying that volume during these attempts — had an effect. But what was interesting is it wasn’t necessarily for the reason one might think. One reason that changing your voice might have an effect is people don’t realize the voice impacts behavior…. Someone says, “You should buy this TV. It’s really great!” I go, “Oh, why are they saying that? They’re trying to persuade me.” But their vocal features are harder for me to detect.

People detected that someone else was trying to persuade them. [But] when we persuade through our voice, not just through what we say, using vocal features, someone can’t tell that we’re trying to do it. So, it wasn’t just that it went undetected, but what happened is it made the communicator seem more confident.

When people were trying to persuade others, they modulated their voice by changing the volume and changing the variation in their volume in a way that made them sound more confident. And because they seemed more confident in what they were saying, that confidence led to greater persuasion.

It’s not so much that people don’t realize that we’re trying to persuade them. They might realize it, but we sound more confident in some cases when we use the right vocal features. And because we sound more confident, we’re more impactful.

Knowledge@Wharton: You noted that this research applies in many arenas. I imagine politics would be one. What are the key takeaways for those who are interested in your research?

Berger: I think the first thing to think about is that we often think about what we say, but we pay a lot less attention to how we say it, whether we’re a politician trying to convince a large audience, whether we’re a leader trying to convince an organization, or a doctor trying to convince a patient, or whether we’re a marketer or a salesperson trying to convince a customer or client. We spend a lot of time on the words we’re saying, and even sometimes think about speaking slowly and standing up straight. We think a lot less about the vocal features. It’s often an unconscious thing that we engage in, and it’s an important area to consider.

There’s a lot of research showing the power of voice just in general. I think in today’s day and age, we’re so used to shooting off an email or texting someone or doing something else because it’s easier. We don’t want to take the time and set up a phone call. We worry about, “Oh, what are they going to say, and how are we going to react?” In this more asynchronous method, written communication, we have more time to construct and refine what to say, so we think we can pitch things better that way.

But, first of all, it’s easier to ignore those pitches. Second of all, some other research shows that the voice is really humanizing. It really brings out the people behind what is being said, and in a wide variety of domains can have a bigger impact. There’s really a power of voice that we often don’t think very much about. While in some cases it’s more effortful to call someone or get together in person, to actually use our voice beyond just the words we say, we should think more about communication modalities and think about which ones are going to be most effective for the impact that we’re trying to have.

Knowledge@Wharton: In a working environment, should this persuade us to walk down the hall and talk to someone, as opposed to sending an email?

Berger: Certainly. There’s a lot of research that came out when email began, showing that emails are easily misinterpreted. We think we’re being very clear, and someone on the other end thinks we’re screaming at them, for example. It’s easy to misinterpret words without some of the vocal cues that come along with them. But in addition to the misinterpretation, I think the voice just has a powerful humanizing effect. It’s much easier to see people as not real people when they’re just words in front of us. Whereas when we hear their voice, we really get that richness, that sense of who they are as an individual. They have a sense of mind. We may treat them more fairly, and we may react better to them because we hear that voice.

Not only should we walk down the hall to talk to a colleague, but maybe we want to do that pitch in person or over the phone, rather than just over email. We should think about situations where that humanizing power of voice is useful, where the persuading power of voice is useful, and make sure to use those features of our voice that can help us.

Knowledge@Wharton: Where do you think this research will take you next?

Berger: This is my first foray into this space. My colleague, Alex Van Zant, who is the first author on this paper, knows a lot about voice, but we’re doing a bunch of work on analyzing text and on analyzing paralinguistic cues or vocal features.

Another project we’re doing, for example, is looking at customer service calls. [We’re looking at how] the words agents use might make customers more satisfied, but also how they say those words. Think about pauses. We’re having a conversation. Should I pause more or less? Maybe pausing as a customer service agent gives time for the customer to respond. “Yes, yes, oh yes — I agree with what you’re saying.” And that agreement helps facilitate that interaction.

What I find particularly interesting about this area as a scientist, but also as someone who teaches in a business school, is that we often don’t think about the words that we put out there in the world. But words are really powerful. We tell our kids there are certain magic words that have more impact than others. Words do a couple of things. One, they reflect something about the people who create them, so you can learn a lot about a person from the words they use. In an employee context, for example, you can tell whether someone is going to stick around at a company, get fired or leave based on the words they use in their email. Whether those words are more similar to or different from the language that the company tends to use can give a signal of whether they’re likely to stick around.

But words not only reflect things about the people who create them, they also impact the people who consume them, whether they’re individuals or the culture more broadly. In the case of persuasion, we have an audience that we’re trying to persuade. But more generally, whether we’re writing a song, creating a movie or doing anything that involves language, thinking about how those words impact that audience is a really powerful opportunity. We’re both studying the words themselves, as well as how those words are used for insight into human behavior.